PMUTs-based Ultrasound Proximity detection for Robot robotic Tactile Sensing

In the context of compliant robotic grasping and safe physical human-robot interaction, accurate pre-contact approach speed sensing in real time at low speeds and short distances remains a critical challenge. This paper presents a PMUT-enabled ultrasonic Doppler sensing scheme that directly estimates dynamic approach speed via frequency-shift extraction, overcoming the limitations of conventional ranging-based methods. The proposed PMUT sensor operates in conjunction with a power spectral density (PSD)-based processing algorithm to achieve high-precision speed detection under short-distance (5-20 cm) and low-speed (0.01-0.1 m/s) conditions - a regime where traditional Fourier-transform techniques exhibit significant errors. Experimental result shows that the system maintains a detection error below 1% across the target speed range, with a latency of approximately 20 ms. Taking the advantage of enabling real-time, low-latency speed perception in close-proximity scenarios, this work provides both a hardware platform and an algorithmic foundation for embedding dynamic proximity sensing into robotic tactile systems, thereby facilitating safer and more responsive human-robot interaction through predictive control.